CORE 2 (Driving Biology Projects) HIGH LEVEL DESCRIPTION Notwithstanding their current utility, i2b2 and SHRINE lack the ability to perform temporal analyses which are needed to discover seasonal trends, cause-and-effect relationships, time dependencies of treatment options, and temporal effects of treatments. As an example, it would be obviously useful to have a temporal analytic i2b2 cell that would allow an invesfigator to issue an i2b2 query that asks "find all patients with diabetes and retinopathy, where the diabetes was noted before the retinopathy." Currently, the i2b2 Workbench can only identify the patients with both diagnoses but not the sequence in which they were noted. To do othenwise requires a SQL-sawy applications programmer and creates a serious bottleneck and tough workarounds for clinical researchers (unassisted by programmers) seeking to investigate comparative effectiveness, or pharmacovigilance where the ability to readily dissect out abstractions of episodes and abstractions of precedence and overiapping and quantifications thereof becomes essential. Prior efforts in this domain fall Into two broad categories: creation of a sophisticated internal database representation of time (including time intervals), with an equally sophisticated temporal query language and database management system and use of a simple, time-stamp-based database architecture coupled with a sophisticated temporal query language. (27, 28). The goal of this proposal is to achieve a significant breadth of temporal expressiveness using existing time-stamped databases (29). In working toward this goal, we leverage work over the last twenty years that permits a pragmatic hybrid ofthe two approaches mentioned above.